Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!zaphod.mps.ohio-state.edu!cs.utexas.edu!uunet!enterpoop.mit.edu!ai-lab!hal.gnu.ai.mit.edu!not-for-mail
From: mycroft@hal.gnu.ai.mit.edu (Charles Hannum)
Newsgroups: comp.os.386bsd.bugs
Subject: Re: Gcc 2.3.3 bug ?
Date: 6 Apr 1993 15:21:16 -0400
Organization: dis
Lines: 34
Message-ID: <1psl7c$1pj1@hal.gnu.ai.mit.edu>
References: <1993Apr6.154454.22346@cm.cf.ac.uk>
NNTP-Posting-Host: hal.ai.mit.edu
In article <1993Apr6.154454.22346@cm.cf.ac.uk> paul@isl.cf.ac.uk (Paul)
writes:
>
> main()
> {
> long int Domain_Class[253][229][2];
> }
You are obvious overflowing your process's stack limit.
> 0x71 in main () (breaks.c line 2)
> 2 (breaks.c)
> (gdb) disass
> Dump of assembler code from 0x68 to 0x78:
> 0x68 <main>: pushl %ebp
> 0x69 <main+1>: movl %esp,%ebp
> 0x6b <main+3>: subl $0x71288,%esp
> 0x71 <main+9>: call 0xf8
> 0x76 <main+14>: leave
> 0x77 <main+15>: ret
> End of assembler dump.
Since the function does nothing, it would seem like the step of setting
up the variable on the stack should be omitted. In this case, the
`defer-pop' optimization and the implicit call to __main() (for C++
constructors) are getting in the way.
It's not clear that this is really a bug in GCC.
--
\ / Charles Hannum, mycroft@ai.mit.edu
/\ \ PGP public key available on request. MIME, AMS, NextMail accepted.
Scheme White heterosexual atheist male (WHAM) pride!